Research Scientist for Speech Synthesis

Research Scientist, Text-to-Speech

Oddin.gg

  About Valka.ai   Valka, a visionary spin-off from the Realms Group (the parent company of Oddin.gg), is on a mission to revolutionize the way people create and experience digital content.   Our team believes that content shouldn’t just be consumed; it should be co-created in real time, blurring the lines between imagination and reality. By harnessing the power of cutting-edge AI, we aim to build an interactive human-digital platform where virtual characters respond dynamically to each user’s voice, text, gestures, and more.   This is your chance to join a diverse group of innovators who are driven to redefine what’s possible in generative content. Together, we’re changing the paradigm from passive viewing to active participation, unlocking new creative frontiers across gaming, entertainment, education, and beyond.

Research Scientist - Model Team

Mirelo

Mirelo AI is building the next generation of creative tools by generating realistic sound, speech and music from video. We develop cutting-edge foundational generative AI models that "unmute" silent video content and create custom, hyper-realistic audio for gaming, video platforms, and creators. Our technology empowers global storytellers to transform their content. We recently closed a $41 million Seed round co-led by Andreessen Horowitz and Index Ventures with participation from Atlantic, and are rapidly expanding across Product, Engineering, Go-to-Market, and Growth. About the Role At Mirelo, you’ll work at the centre of how we build the next generation of multimodal video-to-audio models. This role is deeply hands-on and research-heavy: with a great H100/200-per-engineer ratio you explore and build new multimodal models and push the boundaries of what’s possible in music, sound, and speech generation. You’ll collaborate closely across research and engineering, run focused ablations, and translate experimental results into clear next steps for the team. From data curation to deployment, you’ll help shape the full lifecycle of the models that power our products and partnerships. KEY RESPONSIBILITIES * Design, implement and train large-scale multimodal generative models for audio generation (diffusion and/or autoregressive models). * Explore new modeling ideas for audio generation (music, sound, speech) while taking inspiration from the language and image domains. * Develop and experiment with post-training for new capabilities (fine-grained control, in/out-painting, editing, …) * Conduct rigorous ablation studies, get actionable insights and communicate results to the team to discuss new research directions. * Contribute hands-on to all stages of model development including data curation, experimentation, evaluation, and deployment. IDEAL CANDIDATE PROFILE * Hands-on experience in training large-scale generative models in a fast-paced research environment. * Deep understanding of cutting-edge methods and ML research in at least one of the domains: image, language, video or audio (specific audio experience not necessary, but nice to have). * Strong proficiency in PyTorch, transformer architectures, and the full ecosystem of modern deep learning. * Solid understanding of distributed training techniques—FSDP, low precision training, model parallelism * Strong track-record in working on generative models (publications in top-tier venues, open-source contributions or applied ML projects). NICE TO HAVE * Proficiency with profiling, debugging, and optimizing single and multi-GPU operations using tools like Nsight or stack trace viewers. * Strong software engineering skills/experience in collaborating on large codebases that go beyond PhD research code. * Experience with generative models for audio (sound, music or speech) and audio codec design. WHY JOIN? * Join at a pivotal moment. We've secured fresh funding and are gaining traction - now is when your contributions can make a real difference to our success. * True ownership from day one. You'll have genuine autonomy and responsibility. Your ideas and work will directly shape our product and company direction. * Competitive compensation and equity. We offer strong packages that ensure you share in the success you help create. * Build for the next generation of creators. Be part of the innovation that will transform how creators work and thrive. We welcome applications from all individuals, regardless of ethnic origin, gender, disability, religion or belief, age, or sexual orientation and identity.

Research Scientist - Audio Codec

Mirelo

Mirelo AI is building the next generation of creative tools by generating realistic sound, speech and music from video. We develop cutting-edge foundational generative AI models that "unmute" silent video content and create custom, hyper-realistic audio for gaming, video platforms, and creators. Our technology empowers global storytellers to transform their content. We recently closed a $41 million Seed round co-led by Andreessen Horowitz and Index Ventures with participation from Atlantic, and are rapidly expanding across Product, Engineering, Go-to-Market, and Growth. About the Role At Mirelo, we’re pushing the limits of what generative audio can do, and our ability to innovate depends heavily on the quality of our underlying audio representations. In this role, you’ll work at the core of our modeling stack - designing, training, and evaluating neural audio codecs that directly shape the performance of our next-generation music and sound models. You’ll collaborate closely with the model team, experiment with both continuous and discrete representations, and build the evaluation tools that help us understand what actually moves the needle. Your work will sit at the foundation of building the best-sounding generative models in the world. KEY RESPONSIBILITIES * Develop and implement new neural audio codecs for sound, music and speech that push the state-of-the art in sound quality and are optimized for the the use case of generative models. * Think about the specific challenges that arise when the codec is primarily used as a latent representation in the context of generative audio models (in the end, the ultimate goal is to build the best audio generative models) * Explore the trade-offs of continuous (as typically used for diffusion models) vs. discrete audio representations (as typically used for autoregressive models). * Develop benchmarking pipelines for codec evaluation * Conduct initial experiments with generative models to verify that a new candidate codec is actually useful for our downstream tasks. IDEAL CANDIDATE PROFILE * Strong background in deep learning for audio: neural codecs, source separation, speech models, or generative audio systems * Specific hands-on experience in designing and training neural audio codecs * Solid understanding of audio signal processing fundamentals * Strong track record (research and/or open-source) in the field of audio ML NICE TO HAVE * Hands-on experience with generative audio models and good intuition of how the choice of the codec influences the training and performance of the generative model * Strong publication record (e.g., NeurIPS, ICML, ICLR, Interspeech, ICASSP, WASPAA) WHY JOIN? * Join at a pivotal moment. We've secured fresh funding and are gaining traction - now is when your contributions can make a real difference to our success. * True ownership from day one. You'll have genuine autonomy and responsibility. Your ideas and work will directly shape our product and company direction. * Competitive compensation and equity. We offer strong packages that ensure you share in the success you help create. * Build for the next generation of creators. Be part of the innovation that will transform how creators work and thrive. We welcome applications from all individuals, regardless of ethnic origin, gender, disability, religion or belief, age, or sexual orientation and identity.